-1

I have a set of objects described by a set of features (that are represented by real numbers).

For a given feature, I would like to identify its peaks, that is, the set of values of high probability and that are local maximums. I think that I would be able to do this by estimating the probability density function of each feature.

Intuitively, I think that we can implement an estimation of the probability density function by using histograms. But I'm not sure if I'm right. And, when we use histograms, we need to define some parameters (bin size, for example). I would prefer some method that needs less parameters.

What are the most common methods for estimating the probability density function and that depends on just a few parameters?

Zaratruta
  • 2,097
  • 2
  • 20
  • 26

1 Answers1

2

You can certainly estimate the probability density function using histograms! However, this brings complications depending on the size of your dataset. For example, where do the bins begin?

Have you heard of KDE estimation? There is a lot to talk about here but these two links below should provide you a lot of insight:

Sarthak Batra
  • 21
  • 4